Drop Test Simulation of a BGA Package: Methods & Experimental Comparison

Transcription

1 Drop Test Simulation of a BGA Package: Methods & Experimental Comparison Chris Cowan, Ozen Engineering, Inc. Harvey Tran, Intel Corporation Nghia Le, Intel Corporation Metin Ozen, Ozen Engineering, Inc.

2 SOLID MODEL Generalized BGA Package Substrate / Die / etc. Solder ball PC board

3 EXPERIMENTAL STANDARD JEDEC* Standard JESD22-B110A Standardized mechanical shock acceleration to reduce experimental variation Half-sine pulse Input-G method Table from JESD22-B110A * Joint Electron Device Engineering Council (JEDEC Solid State Technology Association)

4 EXPERIMENTAL STANDARD JEDEC Standard JESD22-B111 Standardized Experimental Drop Test Board 15 components compare performance at 6 unique board locations PCB Young s Modulus: 20 GPa +/-2GPa Mount PCB to drop test fixture with screws through 4 holes Horizontal board orientation with components facing down to maximize board flexure (primary failure mode) Figure from JESD22-B111

5 BOUNDARY CONDITIONS Fixture attachment: Locate fixture attachment point as specified in JESD22- B111. Considered an ideal connection: no friction, no sliding Set X,Y,Z-displacement = 0 (excluding displacement and force method)

6 TIME STEP Guidelines for Integration Time Step (ITS) 1 Set ITS small enough to resolve highest mode that contributes to the response Using approximately 20 points per cycle of highest frequency of interest results in reasonably accurate solution for the Newmark method ITS = 1 / (20f) Example Mode Frequency (Hz) Critical Time Step (s) Total Time (s) Total # Steps Sample Run Time E min E min E hours E hours * Run Time: 3 GHz PentiumD, 2 GB RAM, 15,000 elements, 17,000 nodes

7 PROCEDURE SUMMARY BGA DROP TEST Static Analysis Modal Analysis Dynamic Analysis Force Displacement Acceleration Full Time Integration Mode Superposition Reduced HHT Newmark

8 LOADING A variety of loading methods Acceleration Static or Time Transient Input-G standard loading vs. Experimental Fixture displacement = 0 Applied as body load on package A(t) = P*sin(πt/d) A(t) = acceleration (m/s 2 ) t = time (s) d = duration (s) P = peak acceleration (m/s 2 )

9 LOADING Loading continued Displacement D(t) = A(t) Boundary V o =0 and D o =0 Applied at fixture location Complete package motion Acceleration Velocity Displacement Force Add very large mass element to fixture location Very Large = Package Mass x 1E5 Force = Total Mass x Acceleration Complete package motion 0.0E E E E E E E-04 time (s)

10 DAMPING Damping coefficients calculated based on % of critical damping Rayleigh damping coefficients (α,β) ξ i = α/2ω i + βω i /2 where: ξ i = critical damping ratio α = Rayleigh mass damping coefficient β = Rayleigh stiffness damping coefficient ω i = natural circular frequency = 2π*f i f i = mode frequency i = mode number ANSYS Inc. Theory Reference

11 Experimental Comparison Objective: Compare strain & acceleration results from experimental board level measurements to ANSYS simulation 2003 ANSYS, Inc.

12 GEOMETRY

13 GEOMETRY

14 GEOMETRY Bottom side, viewed from top

15 GEOMETRY

16 SIMULATION SETUP ¼ symmetry model BGA package approximated with bulk material Material properties determined through modal analysis method 1% damping Strain & acceleration simulation results taken at single node nearest to sensor location ANSYS implicit solver

17 BOUNDARY CONDITIONS ¼ symmetry boundary conditions Fix UX,UY,UZ at exact fixture locations

18 MODAL ANALYSIS 1 Frequency (Hz) Period (s) Cumulative Mass Fraction (z-direction) Mode Mode Mode 3 Experimental data shows primary oscillating frequency = Hz. 0.1% Difference

19 LOADING CONDITIONS Loads as recorded on testing table Trapezoidal shock profile 50 G, 11ms

20 ACCELERATION Acceleration vs. Time Acel (G) time (s) ANSYS_Acel1 Acel1_Experimental Acel2_Experimental Table_Input_Experimental Acceleration = Second Derivative UZ + Table Input

21 STRAIN ROSETTE #1 Principal Strains, Rosette #1 3.0E E E-04 Strain 0.0E E E-04 time (s) Ros1_Emax Ros1_Emin ANSYS Ros1 1P ANSYS Ros1 3P Calculate Principal Strains (Plane Stress)

22 STRAIN ROSETTE #2 Principal Strains, Rosette #2 3.0E E E-04 Strain 0.0E E E E E-04 time (s) Ros2_Emax Ros2_Emin ANSYS Ros2 1P ANSYS Ros2 3P Calculate Principal Strains (Plane Stress)

23 STRAIN ROSETTE #3 Principal Strains, Rosette #3 2.0E E-04 Strain 0.0E E E E-04 time (s) Ros2_Emax Ros2_Emin ANSYS Ros3 1P ANSYS Ros3 3P Calculate Principal Strains (Plane Stress)

24 ERROR, ACCELERATION 11% Acel1 6% Acel2 2% avg First Peak, During Loading Acceleration vs. Time Fourth Peak, After Loading 6% Acel1 4% Acel2 3% avg Acel (G) time (s) ANSYS_Acel1 Acel1_Experimental Acel2_Experimental Table_Input_Experimental Difference = Experimental Simulation Experimental

25 ERROR, ROSETTE #1 44% First Peak (+) During Loading 1 st Principal Principal Strains, Rosette #1 3.0E E E-04 Strain 0.0E % First Peak (+) -1.0E-04 During Loading 3 rd Principal -2.0E-04 time (s) Ros1_Emax Ros1_Emin ANSYS Ros1 1P ANSYS Ros1 3P Fourth Peak (-) After Loading 1 st Principal Fourth Peak (-) After Loading 3 rd Principal 6% 34%

26 ERROR, ROSETTE #2 Principal Strains, Rosette #2 5% Fourth Peak (+) After Loading 1 st Principal 45% First Peak (-) During Loading 1 st Principal Insignificant: within noise range of strain gages Strain 3.0E E E E E Insignificant: within noise range of strain gages -2.0E E-04 Fourth Peak (+) After Loading 3 rd Principal 48% -4.0E-04 7% First Peak (-) During Loading 3 rd Principal time (s) Ros2_Emax Ros2_Emin ANSYS Ros2 1P ANSYS Ros2 3P

27 ERROR, ROSETTE #3 Principal Strains, Rosette #3 13% Fourth Peak (+) After Loading 1 st Principal 72% First Peak (-) During Loading 1 st Principal 2.0E E-04 Insignificant Strain 0.0E Insignificant -1.0E E-04 Fourth Peak (+) After Loading 3 rd Principal 65% -3.0E-04 11% First Peak (-) During Loading 3 rd Principal time (s) Ros2_Emax Ros2_Emin ANSYS Ros3 1P ANSYS Ros3 3P

28 SUMMARY Large strain gradients at rosette locations Rosette #1 is located on opposite side, dimensioned from chip, possible error source Rosette #1 difference (44%) is same magnitude as location difference of 1/8 inch on gradient Accelerometer 1,2 (avg) Rosette #1 Rosette #2 Rosette #3 During Loading 2% 44% 7% 11% After Loading 3% 34% 5% 13% Oscillating Frequency 0.1%